Simulation of Wear Processes in LS-DYNA

Simulation of Wear Processes in LS-DYNA

Thomas Borrvall, Anders Jernberg and Mikael

Schill, DYNAmore Nordic AB

Liang Deng and Mats Oldenburg, Luleå Technical

University

Motivation and Example

■ Hot forming process reduces life length of tools

■ Hot blank is formed and subsequently cooled (quenching)

■ High contact pressures and cyclic temperatures

■ Scratches due to sliding wear along radii

■ Significant cost incurred in replacing worn out tools

■ Important to understand the mechanisms behind wear

■ A dog bone wear test illustrates

Upper tool

Wear after

several strokes

Wear Simulation of Dog Bone

■ Possible to post-process wear in LS-PrePost

■ Standard wear law or user defined

■ Iteratively modify geometry to simulate long term use

■ Several LS-DYNA runs with intermediate processing in LS-PrePost

Wear from

simulation

Modified tool

geometrySimulation

Tutorial – wear in deep draw of spherical cup

In particular interested in the wear along the radii of the die

Archard’s law

■ LS-DYNA computes the wear as a nodal quantity which is expressed in rate

form as

𝑤 = 𝐾𝑝𝑣

𝐻

■ 𝑤 is nodal wear depth in direction of the surface normal. Surface normal is

computed from average of element normals. Shell element normals in a wear

interface must to be consistently oriented.

■ 𝐾(𝑝, 𝑣) is a dimensionless scale factor which can be given as a function of

contact pressure 𝑝 and sliding velocity 𝑣.

■ 𝐻(𝑇) is the hardness for the contact side. The hardness can be given as a

function of nodal temperature 𝑇.

■ 𝑝 is the nodal pressure.

■ 𝑣 is the sliding velocity.

Add wear to an LS-DYNA input deck

■ Use any of these contacts■ *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE_{MORTAR}

■ *CONTACT_FORMING_SURFACE_TO_SURFACE_{MORTAR}

■ *CONTACT_AUTOMATIC_SINGLE_SURFACE_MORTAR

■ *CONTACT_FORMING_ONE_WAY_SURFACE_TO_SURFACE

■ Set SPR and/or MPR to ”1” on the contact for interfaces of interest

■ Define *CONTACT_ADD_WEAR

■ Point to contact CID

■ WTYPE=0 for Archard’s law

■ P1-P3 parameters (𝐾,𝐻𝑠, 𝐻𝑚)

■ Set NWEAR>0 on *DATABASE_EXTENT_INTFOR

■ NWEAR=1 for wear depth

■ NWEAR=2 for wear depth and sliding distance

*CONTACT_ADD_WEAR

Card 1 1 2 3 4 5 6 7 8

Variable CID WTYPE P1 P2 P3 P4 P5 P6

Type I I F F F F F F

Default None 0 0.0 0.0 0.0 0.0 0.0 0.0

Post process wear in LS-PrePost

■ Run the input in LS-DYNA with s=intfor on the command line

■ Open the intfor file in LS-PrePost

■ In the fringe menu, wear depth and wear sliding dist. are found

Process wear in LS-DYNA

■ Set SPR and/or MPR to ”2” on the contact, this will also make the side available to the intfor file

■ Set NCYC>0 on *INTERFACE_SPRINGBACK_LSDYNA to get *INITIAL_CONTACT_WEAR data written to dynain

■ Each card corresponds to a wear increment for a given node for 1 cycle

■ It is assumed that the results are valid for 1 stage, i.e., NCYC cycles

The intfor file is for assessing the influence

of wear for a given tool geometry, but to

understand how wear affects the process in

the long haul (thousands of repetitions) we

need to import the wear information back

into LS-DYNA*INITIAL_CONTACT_WEAR

Card 1 1 2 3 4 5 6 7 8

Variable CID NID WDEPT

H

NX NY NZ ISEQ NCYC

Type I I F F F F I I

Default None None None None None None None None

1 LS-DYNA sim = 1 wear cycle

NCYC cycles = 1 wear stage

Rerunning with wear information

■ Include the dynain file as is to the original input

■ The dynain should not contain anything but the *INITIAL_CONTACT_WEAR cards

■ Option A – No intermediate processing, rerun the file in LS-DYNA

■ Each node subject to wear will be moved by the wear depth in the direction of wear,

times NCYC, thus completing 1 stage

■ The user is ”blind”, difficult to obtain a reasonable geometry change

■ Option B – Intermediate processing, open the file in LS-PrePost

The LS-PrePost Wear Interface

■ An *INITIAL_CONTACT_WEAR card gives

information from one cycle, now the

intermediate processing step amounts to

determine the wear from one stage

■ Option 1

■ Leave Scale factor curve ID blank






■ Option 1


■ Set a Max wear distance, corresponding to how much

geometry change you allow based on the latest run






■ Option 1




■ Click Compute, LS-PrePost will determine the number

of cycles required for the max wear at any node to

reach Max wear distance, assuming the wear in each

cycle is constant





determine the wear from many cycles

■ Option 1







cycle is constant

■ Repeatedly click Smooth to smooth the geometry

change, to even out local ”spots” in the wear





determine the wear from many cycles

■ Option 1







cycle is constant



■ The geometry change can and should be previewed by

checking Preview throughout






■ Option 1







cycle is constant



■ The geometry change can and should be previewed by

checking Preview throughout

■ If anything goes wrong, set a new Max wear distance

and repeat the procedure, click Accept when satisfied


■ Sometimes the surface hardness changes with

depth, for which the wear can not be assumed

constant in each cycle

■ Option 2

■ Set Scale factor curve ID to a curve containing a scale

factor as function of total wear depth d

■ The procedure is then follows the one of Option 1

■ After the geometry change is accepted, save

the file to a new input for rerunning in LS-DYNA

with the updated tool geometry

■ This describes one stage in the entire wear

process, to be repeated

Wear after three consecutive wear simulations

Initial geometry

1st run

2nd run

3rd run

Blank holder

Blank

Tool

Fringe plot of total wear

Automated multi-stage run script

■ For cases when the intermediate processing step can be assumed identical

everything can be done using a script

#!/bin/bash

mppsub 32 mppdyna_d_dev_abcde_platformmpi -c -l -b i=run.k memory=400m

lspre43 –nographics c=wear.cfile

Clean



Clean



# Don't Clean. keeping the results from last run

# All previous wear results are now saved in "run.k”

#Note that the submit script, mppsub in this case, must not return the prompt until the

#simulation has finished completely.

#The ”original_input.k” is copied to a ”run.k” before this script is started.

run.sh

LS-PrePost command file

■ Read in the original keyword file

■ Import the dynain file

■ Set the max wear distance to 0.5

■ Compute the wear

■ Smooth the wear once

■ Update the node coordinates

■ Write a new input file, run.k

openc keyword "/disk/home/anders/wear/original_input.k”

import keyword nooffset

import keyword "/disk/home/anders/wear/dynain”

save keywordoutversion 7

wear maxdist 0.5

wear compute

wear smooth

wear accept

save keyword "/disk/home/anders/wear/run.k”

exit

wear.cfile

Concluding remarks and general recommendations

■ LS-DYNA and LS-PrePost can be used in parallel to simulate wear processes

■ ”In the beginning” but conceptually works well

■ Manual or automated processing

■ For the best results

■ The contact pressure and friction should be smooth

■ Mesh density should be relatively fine for parts where wear is important

■ Stiff contact will localize wear for faceted geometries, soften to distribute

■ One wear stage must not change the geometry too much as this

will result in unrealistic wear in subsequent stages

■ Part is available in R9.0 and all is available in R10

■ R9.0 – post processing only

■ R10 – entire process simulation

Thank you!

Your LS-DYNA distributor and

more

■ LS-DYNA

■ More contact types(?)

■ More advanced wear laws(?)

Future work – mainly driven by user feedback

■ LS-PrePost

■ Local smoothing(?)

■ Manual interaction(?)

Simulation of Wear Processes in LS-DYNA

Documents

Transcript of Simulation of Wear Processes in LS-DYNA